Gilquin 2008 Combining Contrastive and Interlanguage Analysis Draft
Total Page:16
File Type:pdf, Size:1020Kb
Combining contrastive and interlanguage analysis to apprehend transfer: detection, explanation, evaluation Gaëtanelle Gilquin Centre for English Corpus Linguistics F.N.R.S. – Université catholique de Louvain Abstract This paper shows how contrastive analysis and interlanguage analysis can be combined into a model which aims to detect phenomena of transfer, explain them and evaluate them in terms of pedagogical relevance. The model, referred to as the Detection-Explanation- Evaluation (DEE) transfer model, relies on Granger’s (1996) Integrated Contrastive Model and Jarvis’s (2000) unified framework for transfer research, and uses corpus data as a basis for a range of comparisons between L1 (mother tongue), L2 (target language) and IL (interlanguage). As an illustration, the use of even if by French-speaking learners is investigated, and the possible influence of L1 is examined in the light of the different comparisons involved in the model. Finally, some limitations of the DEE transfer model are discussed, all of which point to possible developments that could be implemented in the future. 1. Introduction The importance of transfer in second language acquisition can now be regarded as an established fact. As observed by Kohn (1986: 21) two decades ago already, “there is no doubt that, despite its sometimes irritatingly elusive character, transfer is one of the major factors shaping the learner’s interlanguage competence and performance”. Two main sources of data have been used over the years to investigate the influence of the mother tongue (L1) on the interlanguage (IL), namely contrastive analysis (CA) data and learner data. Selinker (1989: 287) notes that “CA is the best place to begin language transfer studies since structural congruence (or at the least, partial structural similarity) is most probably necessary, though not sufficient, for most types of language transfer to occur”. Indeed, Schachter’s (1974) analysis, which investigated the influence of L1 on the use of English restrictive relative clauses by learners from several mother tongue backgrounds, was demonstrated to be flawed, for lack of reliable CA data (Zhao 1989, Kamimoto et al. 1992). Without samples of learner production to rely on, however, studies of transfer, as faultless as they may be from a contrastive point of view, could offer nothing but useless predictions. Both contrastive analysis and interlanguage analysis are therefore necessary for a sound and systematic assessment of the role of transfer in second language acquisition.1 2 Gaëtanelle Gilquin In this paper, two models of transfer which combine a CA and an IL approach, Granger (1996) and Jarvis (2000), are compared in an attempt to develop a refined method to apprehend transfer. Three varieties of language will be considered, viz. L1, L2 (target language) and IL, and will lead to a number of comparisons which aim to detect transfer (when is it at work?), explain it (what are its origins?) and evaluate it from a pedagogical perspective (is it pedagogically relevant?). Reflecting this threefold purpose, the model will be referred to as the Detection-Explanation-Evaluation (DEE) transfer model. Because the interest here is in phenomena of transfer as they manifest themselves in naturally-occurring language, use will be made of corpora, whose value for linguistic descriptions has been underlined by numerous studies over the last few years. The paper is organised as follows. First, the phenomenon of transfer is briefly described and recontextualised, with special emphasis on its links with contrastive analysis and interlanguage analysis. Granger’s (1996) Integrated Contrastive Model and Jarvis’s (2000) unified framework for the study of L1 influence are then presented and compared. On the basis of this comparison, a new model of transfer is proposed, the DEE transfer model, which serves the threefold purpose of detecting, explaining and evaluating cases of transfer. The model is illustrated by means of a case study of the use of even if by French- speaking learners, after which some of its limitations are outlined. The article ends with some concluding remarks. 2. The transfer mystery After a period in the 1950s and 60s when transfer was seen as all-pervasive and a period in the 1970s when its existence was denied, we have now reached a time when it is “generally accepted that transfer does occur, but is a far more complex phenomenon than hitherto believed” (Benson 2002: 68). Its exact nature, the circumstances in which it occurs or the processes it relies on in the learner’s mind are still, to a large extent, a mystery to linguists. As Dechert & Raupach (1989: ix) put it in their introduction, “[i]n spite of three decades of intensive research […] there is still no generally accepted agreement of what transfer in language actually is”. Lado (1957: 2) refers to the transfer of “the forms and meanings and the distribution of forms and meanings of [individuals’] native language and culture to the foreign language and culture”. Odlin’s (1989: 27) definition is broader, as it covers the influence of “any other language that has been previously (and perhaps imperfectly) acquired”. In other words, transfer is not limited to the influence of L1 on L2, but could involve the influence of, say, L2 on L3 or L3 on L5. Neither of these definitions, however, explicitly raises the possibility of a simultaneous influence of several languages. Nor do they allow for the fact that a language, most typically the native language, may be influenced by subsequently acquired languages. In order to take such phenomena into account, transfer will be defined here in very general terms, as the influence, within an individual’s Combining contrastive and interlanguage analysis 3 linguistic system, of one or more languages over another.2 Moreover, we will consider both cases of negative transfer, when the influence leads to erroneous usage,3 and positive transfer, when it leads to correct usage. The notion of transfer is closely related to contrastive analysis. In fact, as pointed out by Ringbom (1995: 581), the original idea behind contrastive analysis was to bring to light the problems that learners from a specific mother tongue background might experience in learning a given foreign language. However, the study of learner language (so-called “interlanguage”, see Selinker 1972), initially investigated within the frame of error analysis, revealed that problems could occur in cases not predicted by contrastive analysis and, conversely, that not all areas predicted as difficult by contrastive analysis were actually problematic for learners (Ringbom 1995). Thus, while both contrastive analysis and interlanguage analysis appeared necessary to make discoveries about transfer, it also turned out that the link between the two was far from obvious and required further research. The model of transfer presented here, the DEE transfer model, relies on two models combining contrastive and interlanguage analysis, namely Granger (1996) and Jarvis (2000), and aims to provide a systematic method for detecting transfer, explaining it and evaluating its pedagogical relevance. It involves three 4 varieties of language (L1, L2 and IL) and six types of comparison, as summarised in Table 1, which also shows the abbreviations that will be used throughout the article. Of these six comparisons, the first three are instances of contrastive analysis, whereas the last three involve so-called contrastive interlanguage analysis.5 Each of them serves a particular (primary) purpose, namely detection for NL1/IL and ILa/ILb…ILz (see Section 6.1), explanation for OL1/OL2 and SL1/2/TL2/1 (see Section 6.2) and evaluation for NL2/IL and ILa/ILa (see Section 6.3). Table 1. Types of comparison and their abbreviations Abbreviation Explanation Example for French learner of English OL1/OL2 Comparison of original L1 and Original French vs. original original L2 English SL1/2/TL2/1 Comparison of source L1 and Source French vs. translated translated L2 or vice versa English or source English vs. translated French NL1/IL Comparison of native L1 and Native French vs. learner IL English NL2/IL Comparison of native L2 and Native English vs. learner IL English ILa/ILa Comparison of interlanguage French learner English vs. by learners from the same L1 French learner English ILa/ILb…ILz Comparison of interlanguage French learner English vs. by learners from different L1s Spanish learner English vs. German learner English, etc 4 Gaëtanelle Gilquin Although the model is mainly illustrated through cases of transfer from L1 French to L2 English, and most of the examples concern lexical (and some syntactic) phenomena, it is possible to apply the DEE transfer model to other languages, other types of transfer – e.g. from L2 to L3 (Hammarberg 2001: 23) or from L2 to L1 (Pavlenko & Jarvis 2002) – and other phenomena (phonological, pragmatic, etc).6 And while all the illustrations come from written corpus data, the model could also serve to study transfer in speech, provided the necessary material is available. 3. Granger’s (1996) Integrated Contrastive Model In Granger’s (1996) Integrated Contrastive Model (ICM), two components are combined in an attempt to shed light on the phenomenon of transfer. The contrastive analysis (CA) component compares (i) original data from one language with original data from another language and (ii) source language and its translation in another language. The contrastive interlanguage analysis (CIA) component, similarly, consists of two types of comparison: (i) native language and a non-native variety of this language and (ii) different non-native varieties of one and the same language. The peculiarity of the model is that it exclusively relies on authentic data coming from computerised corpora. This has not always been the case in contrastive and interlanguage studies. The first comparisons of two or more languages were often intuition-based and the traditional error analyses of learner language usually relied on very small collections of texts.7 Recently, however, with the computer revolution, large amounts of machine-readable authentic data have become available.